Validity and reliability of food frequency questionnaire used in age-related eye disease studies in Turkish adults

Newly developed dietary assessment tools for lutein and zeaxanthin are correlated with 24-hour diet recalls, but are not a valid measure of intake in Australian and United Kingdom adults

Habitual dietary intake measurement of carotenoids lutein and zeaxanthin (L/Z) has often been omitted or attempted with tools of unknown validity in past research. It was hypothesized that the dietary assessment tool, the L/Z screener, developed as part of this study, would be valid with agreement within 0.25 mg/day when compared against multiple 24-hour diet recalls in healthy Australian and United Kingdom adults. Two screeners with 91 food items were developed, 1 with a recall timeframe of a month and the other a week. Over 4 weeks, 56 Australian and 47 United Kingdom participants completed 4 weekly screeners, 2 monthly screeners, and eight 24-hour diet recalls. Validity was assessed through Bland-Altman plot analysis. L/Z intake measured by all tools was significantly correlated, with correlation coefficients from 0.58 to 0.83. Despite these correlations, the screeners were not valid, with poor Bland-Altman plot agreement when compared with the diet recalls. The Australian weekly screener performed best, demonstrating a mean difference of 0.51 mg/day and 95% limits of agreement between -1.46 mg/day and 2.49 mg/day of L/Z intake. Baby spinach, broccoli, and pumpkin provided the greatest proportion of L/Z intake. The low validity may be explained by high rates of misestimation or missed capture of moderate to high L/Z containing foods such as baby spinach. Prior research reliant on correlational statistics for L/Z tool validity should be interpreted with caution, and future screener development should prioritize accurate capture of high contribution foods.